Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository

Page created by Amber Singh
 
CONTINUE READING
Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
CENTRE FOR NEUROSCIENCE IN EDUCATION, UNIVERSITY OF CAMBRIDGE

              Understanding
            Mathematics Anxiety
   Investigating the experiences of UK primary and
              secondary school students
                                        Report authors:

                                         Emma Carey1
                                       Dr. Amy Devine1,4
                                        Francesca Hill1
                                       Dr. Ann Dowker3
                                       Dr. Ros McLellan2
                                       Dr. Denes Szucs*1

                                          March 2019

Author affiliations: (1) Department of Psychology; University of Cambridge, UK; (2) Faculty of
Education; University of Cambridge; UK; (3) Department of Experimental Psychology; University of
Oxford; UK; (4) Cambridge Assessment English; Cambridge; UK.
* Correspondence: Denes Szucs; ds77@cam.ac.uk
Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
Table of Contents
  Executive Summary ............................................................................................................................ 2

            1. Background ........................................................................................................................ 2

            2. Aims ................................................................................................................................... 2

            3. Methodology...................................................................................................................... 2

            4. Summary of Key Findings ................................................................................................... 3

            5. Conclusions ........................................................................................................................ 4

            6. Recommendations ............................................................................................................. 4

            7. Acknowledgements............................................................................................................ 5

  The need for this research .................................................................................................................. 6

            8. Maths in the UK ................................................................................................................. 6

            9. What is maths anxiety? ...................................................................................................... 6

  Aims .................................................................................................................................................... 8

  Methods .............................................................................................................................................. 8

  Key Findings: In Full .......................................................................................................................... 10

            10. Validating the modified Abbreviated Mathematics Anxiety Scale (mAMAS) .................. 10

            11. Exploring the relationship between mathematics anxiety and performance ................. 14

            12. Relationship between maths anxiety and developmental dyscalculia ............................ 17

            13. Maths anxiety: Gender differences, developmental change and anxiety specificity ...... 19

            14. Anxiety profiles and their relationship with performance .............................................. 22

            15. Working memory, dyscalculia and maths anxiety in Italian students ............................. 26

            16. Qualitative research: Experiences and origins of mathematics anxiety .......................... 28

  General Conclusions ......................................................................................................................... 52

            17. Recommendations ........................................................................................................... 53

  Outcomes .......................................................................................................................................... 55

            18. Publications ...................................................................................................................... 55

            19. Cambridge Science Festival Event.................................................................................... 56

  References ........................................................................................................................................ 57

                                                                                                                                                           1
Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
Executive Summary

Background

The project investigated individuals’ attitudes towards mathematics because of what could be
referred to as a “mathematics crisis” in the UK. Evidence suggests that functional literacy skills
amongst working-age adults are steadily increasing but the proportion of adults with functional
maths skills equivalent to a GCSE grade C has dropped from 26% in 2003 to only 22% in 2011
(National Numeracy, 2014). This number is strikingly low compared with the 57% who achieved the
equivalent in functional literacy skills (National Numeracy, 2014).

While mathematics is often considered a difficult subject, not all mathematics difficulties result from
cognitive difficulties. Many children and adults experience feelings of anxiety, apprehension, tension
or discomfort when they are confronted by mathematics. This may be contributing to a relatively
low level of numeracy amongst UK adults. In this project we investigate primary and secondary
school students’ experiences in the mathematics classroom and beyond, in order to learn more
about the prevalence, nature and resolution of mathematics anxiety

Mathematics anxiety describes feelings of apprehension, tension or discomfort experienced by many
individuals when performing mathematics or in a mathematical context (Richardson & Suinn, 1972).
It has been associated with cognitive difficulties performing mathematical tasks, potentially because
anxiety interferes with our ability to hold and manipulate information in mind (working memory),
but is predominantly an emotional problem (Ashcraft & Krause, 2007). Developmental dyscalculia,
on the other hand, is a cognitive difficulty in acquiring mathematical skills (Devine, Hill, Carey, &
Szűcs, 2017).

Aims

This project had multiple goals. Firstly, we wished to provide estimates of mathematics anxiety
prevalence amongst UK primary and secondary school students. Secondly, we wished to validate an
instrument for measuring mathematics anxiety in this group. Thirdly, we wanted to see how
mathematics anxiety is related to other individual factors, such as maths performance, gender and
individual cognitive differences. Additionally, we used interviews to gain a richer understanding of
students’ experiences with and feelings towards mathematics.

Methodology

Our study had two main phases. In the first of these, we worked with more than 1700 primary and
secondary students to screen for mathematics anxiety, test anxiety and general anxiety, and gain a
measure of mathematics and reading performance. In the second phase, we worked one-to-one
with the children to gain a deeper understanding of their cognitive abilities and feelings towards
mathematics, using a series of sessions administering cognitive tasks, questionnaires and interviews
(Carey, Devine, et al., 2017; Carey, Hill, Devine, & Szűcs, 2017; Devine et al., 2017). Throughout, we

                                                                                                         2
Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
worked on two further sub-projects looking at the relationship between maths anxiety and
performance and cross-cultural comparisons (Carey, Hill, Devine, & Szucs, 2015; Hill et al., 2016).

Summary of Key Findings

o   We have validated, in our large sample of British children, that the modified Abbreviated Math
    Anxiety Scale is reliable (it appears to measure one construct) and valid (this construct seems to
    really be maths anxiety, rather than another form of anxiety or other feelings towards maths
    (Carey, Hill, et al., 2017). See Validating the modified Abbreviated Math Anxiety Scale (mAMAS)
    for more details.
o   We conducted a literature review into the long-established relationship between maths anxiety
    and performance (those with higher maths anxiety tend to have poorer maths performance).
    We conclude that this is likely because anxiety interferes with performance and poorer
    performance increases anxiety, acting as a vicious circle (Carey et al., 2015). See Exploring the
    relationship between mathematics anxiety and performance for more details.
o   In our large sample of British children, we investigated the relationship between maths anxiety
    and developmental dyscalculia. We found that whilst more dyscalculics than typical children
    met criteria for maths anxiety, the majority of those with maths anxiety had normal
    performance (Devine et al., 2017). See The relationship between mathematics anxiety and
    developmental dyscalculia for more details.
o   In a separate group of Italian children, we participated in research looking at developmental
    change, gender differences and specificity of maths anxiety. We found that unlike general
    anxiety, maths anxiety increases with age. The relationship between maths anxiety and
    performance becomes more specific with age – in younger, but not older, children, this
    relationship disappears after accounting for general anxiety. See Maths anxiety: Gender
    differences, developmental change and anxiety specificity for more details.
o   We have identified, in our large British sample, anxiety subgroups. These may increase in
    complexity with age. In our secondary school students, we found that those with anxiety specific
    to academia (high maths and test anxiety) had poorer performance than those with higher, but
    less specific, anxiety. We conclude that this may reflect a dual path in anxiety development and
    maintenance (Carey, Devine, et al., 2017). See Anxiety profiles and their relationship with
    performance for more details.
o   In our smaller subsample of British students, with whom we conducted further testing, we
    looked at the relationship between various cognitive variables and maths performance. It seems
    that a myriad of factors are associated with maths performance, but that basic numerical
    processing is not (unpublished data). See Cognitive factors in mathematics performance for
    more details.
o   In another Italian sample, we investigated specific memory subtypes and their relationship with
    maths anxiety and dyscalculia. Whereas maths anxiety appears to be associated with a deficit in
    verbal working memory and perhaps also visuospatial working memory, dyscalculia is
    associated with deficits in visuospatial memory; both short-term and working memory are
    affected (Mammarella, Hill, Devine, Caviola, & Szűcs, 2015). See Working memory, dyscalculia
    and maths anxiety in Italian students for more details.
o   Our qualitative research has shown that children of 9-10 years are able to discuss their
    experiences and origins of mathematics anxiety. Mathematically anxious children seemed to
    describe negative events with less contextualisation. They were also more likely to discuss

                                                                                                      3
Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
physical sensations in their maths classes and clearly articulated some of the negative
    consequences of maths anxiety. See Qualitative research: Experiences and origins of
    mathematics anxiety for more details.

Conclusions

Each of the completed projects within our study further reveals the complex, multifaceted nature of
mathematics anxiety. It is likely that mathematics anxiety is not a simple construct with only one
cause – rather, it can emerge as a result of multiple predisposing factors including gender, cognitive
abilities and general predisposition towards anxiety, rumination or panicking under pressure. This
helps to explain why mathematics anxiety is robustly correlated to a small degree with many
constructs (e.g. test anxiety, general anxiety and mathematics ability). We have clearly shown that
emotional and cognitive mathematics problems dissociate and therefore require different
intervention strategies. Our qualitative analysis of structured interviews suggests that children as
young as 9 are experts in their own experiences in mathematics and this can be harnessed to further
understand the thought processes underlying maths anxiety. This brings us closer to design effective
prevention and remediation programs for mathematics anxiety.

Recommendations

o   The 9-item modified Abbreviated Mathematics Anxiety (mAMAS) scale developed by this project
    proved to be a reliable tool for investigating math anxiety in school context.
o   Teachers need to be conscious that individuals' maths anxiety likely affects their mathematics
    performance.
o   Teachers and parents need to be conscious of the fact that their own mathematics anxiety might
    influence student mathematics anxiety and that gendered stereotypes about mathematics
    suitability and ability might drive to some degree the gender gap in maths performance.
o   Hence, for parents and teachers, tackling their own anxieties and belief systems in mathematics
    might be the first step to helping their children or students.
o   With our research showing that maths anxiety is present from a young age and goes through
    significant developmental change, we suggest focusing further research on how maths anxiety
    can be best remediated before any strong link with performance begins to emerge.
o   The qualitative part of our research shows that children are able to verbalise the suffering that
    mathematics anxiety causes them. Our qualitative research also points to several potential
    causes of maths anxiety that could be focused upon by further research.
o   Teacher training should clearly highlight the role of both cognitive and affective factors behind
    maths learning in schools.
o   Policy makers should be conscious that emotional blocks can have substantial impact on learning
    potential.
o   Emotional and cognitive problems require completely different interventions.

                                                                                                         4
Understanding Mathematics Anxiety - Investigating the experiences of UK primary and secondary school students - Cambridge Repository
Acknowledgements

The authors would like to thank all members of the Centre for Neuroscience in Education who have
been involved in this project, from its inception to completion.

In addition, we would like to thank all of the schools who accommodated our busy testing schedule;
this research would not be possible without the goodwill of the schools with which we worked. We
would like to thank all participating students for their candour during interviews and willingness to
complete cognitive testing with us.

The Nuffield Foundation is an endowed charitable trust that aims to improve social wellbeing in the
widest sense. It funds research and innovation in
education and social policy and also works to build
capacity in education, science and social science
research. The Nuffield Foundation has funded this
project, but the views expressed are those of the
authors and not necessarily those of the Foundation. More information is available at
www.nuffieldfoundation.org.

Additional funding was provided by the James S McDonnell Foundation 21st Century Science
Initiative in Understanding Human Cognition - Scholar Award 220020370, but the views expressed
are those of the authors and not necessarily those of the Foundation.

                                                                                                        5
The need for this research

Maths in the UK

We investigated individuals’ attitudes towards mathematics because of what could be referred to as
a “mathematics crisis” in the UK. Whilst functional literacy skills amongst working-age adults are
steadily increasing, the proportion of adults with functional maths skills equivalent to a GCSE grade C
has dropped from 26% in 2003 to only 22% in 2011 (National Numeracy, 2014). This number is
strikingly low compared with the 57% who achieved the equivalent in functional literacy skills
(National Numeracy, 2014).

This reduction in maths competence over time could be driven by a change in people’s attitudes
towards mathematics. For example, many people mistakenly hold the belief that maths is a skill we
are born with, rather than one you can learn. This attitude could lead to demotivation in learning,
because people think that however much effort they put into learning mathematics, they are unable
to alter their fixed mathematical ability (Churchman, 2013).

With 4 in 5 adults in the UK having a low level of numeracy, low mathematical skill is costing billions
to both individuals and society at large (National Numeracy, 2014). Many employers report concern
about their employees’ abilities to ‘sense-check’ numbers – i.e. to ensure that numbers reported are
in a ballpark range of what you would expect – resulting in higher costs (National Numeracy, 2014).
Less numerate individuals also earn less than their numerate counterparts (National Numeracy,
2014). This is especially disturbing given that socio-economic background influences a child’s
mathematics achievement by around 10-20% (National Numeracy, 2014). Therefore, this issue could
contribute to a cycle of poverty and decrease social mobility.

What is maths anxiety?

Whilst some problems in maths are related to difficulty understanding the material, other problems
might stem from emotional difficulties. Mathematics anxiety (also called maths anxiety) is a negative
emotional reaction to mathematics, which can interfere with the ability to perform mathematical
tasks. Maths anxiety has many different manifestations, including emotional - for example, feelings
of apprehension, dislike, tension, worry, frustration or fear, physical - for example, butterflies, racing
heart, struggling to catch your breath or behavioural - for example misbehaving in class, avoiding
maths assignments, not studying maths beyond the minimum expected level (Hembree, 1990).

Maths anxiety has been seen in young school children from around the age of 6 (Beilock, Gunderson,
Ramirez, & Levine, 2010; Krinzinger, Kaufmann, & Willmes, 2009; Thomas & Dowker, 2000; Vukovic,
Kieffer, Bailey, & Harari, 2013). However, negative attitudes towards mathematics and maths
anxiety appear to increase when children reach secondary school age, persisting into post-secondary
education and throughout adulthood (Dowker, Sarkar, & Looi, 2016). It is difficult to define the
prevalence of maths anxiety, because measures of maths anxiety are continuous (i.e. each individual

                                                                                                          6
falls somewhere on a spectrum) with no clear cut-off as to whether an individual is maths-anxious or
not (Devine et al., 2017).

Maths anxiety affects individual wellbeing – for example, some students will dread their maths
lessons or avoid doing their maths homework due to a dislike of experiencing negative emotions
(Dowker et al., 2016; Hembree, 1990). There is also a relationship between maths anxiety and maths
performance: individuals who have higher levels of maths anxiety tend to do worse in maths tests
(Ashcraft & Krause, 2007; Carey et al., 2015; Hembree, 1990; Ma & Xu, 2004). Importantly, those
affected by higher levels of maths anxiety may develop other negative attitudes towards
mathematics, avoid or drop out of voluntary maths classes, or avoid careers which require
quantitative skills (Hembree, 1990). Since quantitative skills apply to such a broad range of careers,
this may impose a severe limit on the life choices of somebody with high maths anxiety.

Many potential causes have been identified for maths anxiety. These include some environmental
factors, such as:

    o   Negative experiences in class, e.g. doing badly in maths due to poor basic skills (Maloney,
        Ansari, & Fugelsang, 2011);
    o   Teacher characteristics, e.g. women with anxious female maths teachers are more likely to
        become anxious themselves (Beilock et al., 2010);
    o   Parental gender stereotypes, e.g. parents expecting that their daughter will struggle more in
        maths than their son (Tomasetto, Alparone, & Cadinu, 2011).

Additionally, intellectual factors may be involved in maths anxiety. Children with developmental
dyscalculia (a specific deficit in the acquisition of mathematics skills) and other mathematics learning
disabilities have an increased risk of experiencing mathematics anxiety (Passolunghi, 2011;
Rubinsten & Tannock, 2010).

Furthermore, an individual’s other personal characteristics might influence their maths anxiety. For
example:

    o   Gender - girls are more likely to experience anxiety about maths (Hembree, 1990);
    o   Self-esteem, with a lower self-esteem contributing to higher levels of maths anxiety (Abbasi,
        Samadzadeh, & Shahbazzadegan, 2013);
    o   Learning style (Sloan, Daane, & Giesen, 2002); and
    o   Attitude towards maths – those who generally like maths tend to have lower maths anxiety
        levels than those who dislike maths (Hembree, 1990).

Whilst studies have identified each of these things as potentially related to individuals’ levels of
maths anxiety, the direction of the relationship between maths anxiety and any of the above factors
is unclear. For example, in the case of the relationship between maths anxiety and maths
performance, it is unclear whether those with lower maths performance are more likely to become
anxious about maths (supported by Ma & Xu, 2004; Maloney et al., 2011; Meece, Wigfield, & Eccles,

                                                                                                         7
1990), whether maths anxiety acts to lower maths performance (supported by Ashcraft & Faust,
1994; Ashcraft & Krause, 2007; Morsanyi, Busdraghi, & Primi, 2014) or whether the relationship is
reciprocal, with both causal directions being active (Ashcraft, Krause, & Hopko, 2007; Jansen et al.,
2013; Luo et al., 2014).

Aims

The overarching aim of our research was to investigate how maths anxiety impacts upon primary
and secondary school students’ wellbeing and learning. This encompassed several key objectives:

    1. To assess the prevalence of mathematics anxiety in a sizeable UK child sample;
    2. To provide a reliable, valid and easy to use measure of maths anxiety for UK children;
    3. To investigate gender differences in mathematics anxiety;
    4. To investigate the relation of maths anxiety to performance;
    5. To see how maths anxiety relates to other cognitive variables;
    6. To use interviews to shed light on the origins, everyday experience of mathematics anxiety
        and coping strategies for mathematics anxiety.
    7. To raise public awareness of maths anxiety and disseminate our findings.

Methods

We used a mixed methods approach combining quantitative and qualitative data methods.
Quantitative research methods involve measuring people’s performance and attitudes using
instruments such as tests and questionnaires. Qualitative research aims to provide more of an
insight into individual experiences, e.g. allowing participants the chance to tell us their experiences
without being constrained to several limited choices on a questionnaire. Using mixed methods is one
of the key features of our research as it enables us to gain both an understanding of the numbers
(answering questions like: how many people have high levels of maths anxiety and how does maths
anxiety relate to other measurable individual differences?) and a rich understanding of individual
experiences.

Data collection occurred in two main phases. Phase 1 involved screening approximately 1800
primary and secondary school children for mathematics anxiety, in whole classes or year groups.
Screening tests included questionnaires assessing mathematics anxiety, and two other related forms
of anxiety: test anxiety and general anxiety. During screening, students also completed age-
standardised mathematics and reading tests. Participating schools were located in South East
England, comprised a mix of urban and rural schools, and covered a wide range of socio-economic
backgrounds.

Subgroups of children with the highest mathematics anxiety (a “high maths anxiety” group) or
average mathematics anxiety (a control group) were selected to take part in phase 2. Phase 2
involved further assessments of maths anxiety and general anxiety, assessment of the children’s
cognitive performance (IQ, working memory, executive function), as well as two semi-structured
                                                                                                          8
interviews. The interviews probed children’s feelings about mathematics, the origins of these
feelings, their experiences of mathematics in the classroom and coping strategies, as well as the
children’s perceived mathematics competence. Full details of the samples, methods, instruments,
and statistical analyses can be found in our published articles (Carey, Devine, et al., 2017; Carey, Hill,
et al., 2017; Devine et al., 2017).

                             Phase 1                                       Additional Projects
      o    1800 students (half in year 4, half in year 7).                 o   Conducted a review of
      o    Attending schools across South East England.                        the literature
      o    Tested in whole-class or whole-year groups for                      concerning the
           around 2 hours altogether.                                          relationship between
      o    We measured:                                                        mathematics anxiety
               o Maths anxiety;                                                and mathematics
               o Test anxiety;                                                 performance.
               o General anxiety;                                          o   Investigation of
               o Maths performance;                                            mathematics anxiety
               o Reading performance.                                          in a large sample of
                                                                               Italian school
                        Phase 1 Analysis                                       students.
                                                                           o   Investigation of the
      o    Validated our maths anxiety questionnaire, which
                                                                               relationship between
           was an adapted version of an adult maths anxiety                    mathematics anxiety,
           questionnaire.                                                      dyscalculia and
      o    Looked at how maths anxiety, test anxiety and
                                                                               working memory.
           general anxiety cluster in individuals, and how this
           relates to mathematics performance.
      o    Looked at how maths anxiety relates to cognitive
           difficulties with maths (developmental dyscalculia).

       Phase 2: Maths Anxiety Sample                             Phase 2: Random Sample
      o    120 students (half from year 5, half              o    200 students (half from year 5, half
           from year 8).                                          from year 8).
      o    Worked with students one to one for               o    Worked with students one to one
           around 2 hours each (split across 4                    for around 1 hour each (split across
           sessions).                                             2 sessions).
      o    Half had high maths anxiety; half had             o    Students were sampled randomly.
           average maths anxiety.                            o    We measured:
      o    We measured:                                               o IQ (verbal and non-verbal);
               o Mathematics anxiety;                                 o Verbal and visuospatial
               o General anxiety;                                         working memory.
               o IQ (verbal and non-verbal);                          o Visuospatial short term
               o Interviews about origins and                             memory.
                   experiences of maths anxiety.                      o Executive functions.

                                           Phase 2 Analysis
      o    Use interviews to enrich our understanding of the qualitative experience of
           mathematics anxiety.
      o    Investigate the cognitive variables associated with mathematics anxiety and
           performance.

                                                                                                         9
Key Findings: In Full

The following sections describe the details of our investigation.

Validating the modified Abbreviated Mathematics Anxiety Scale (mAMAS)

Why we needed a new child maths anxiety measure
For our project, reliably and accurately measuring maths anxiety in children aged 8-13 years was of
vital importance. Many measures exist for measuring maths anxiety in adults (e.g. Hopko,
Mahadevan, Bare, & Hunt, 2003; Richardson & Suinn, 1972), and several measures of childhood
maths anxiety have also been developed in recent years (e.g. Ramirez, Gunderson, Levine, & Beilock,
2013; Wu, Barth, Amin, Malcarne, & Menon, 2012). Adult measures of maths anxiety are not
suitable for use in children because they often refer to advanced mathematics which children have
not been exposed to. For example, the Abbreviated Maths Anxiety Scale (AMAS) refers to anxiety
elicited by checking the tables in the back of your textbook, something which none of the students in
our sample will have had to do in their maths classes (Carey, Hill, et al., 2017).

On the other hand, childhood maths anxiety scales are also limited. Sometimes this is either because
they have not been adequately tested for reliability (an assessment of whether a test measures just
one construct consistently) or validity (an assessment of whether a test measures the construct
which you want it to). Additionally, some childhood maths anxiety scales are only suitable for use
with a very narrow range of children. For example, the Child Mathematics Anxiety Questionnaire is
only suitable for use during a 2 year period of primary education, because it asks children questions
such as “How would you feel if you were given this problem? There are 13 ducks in the water. There
are 6 ducks in the grass. How many ducks are there in all?” (Ramirez et al., 2013). This question is
likely to elicit much more anxiety in a child younger than the test was designed for (as they may not
be familiar with this kind of problem) and much less anxiety in a child older than the test was
designed for (as they will find the problem simple, and research shows that more complex problems
elicit higher levels of mathematics anxiety). Beyond the issue of age, it is possible that even within a
narrow age range, referring to specific problems might not be ideal. For example, it may confuse the
relationship between mathematics anxiety and maths performance even further: lower performing
children are more likely than high performers to feel anxiety about a specific question which is
above their mathematical capabilities, thus it may appear that there is a stronger relationship with
maths performance and maths anxiety than is really present (Carey, Hill, et al., 2017).

Other tests of maths anxiety in children are the Scale for Early Mathematics Anxiety (Wu et al.,
2012), which also refers to specific questions and therefore is subject to the same problem as the
Child Mathematics Anxiety Questionnaire, and the Mathematics Attitude Questionnaire (Thomas &
Dowker, 2000). The latter does not refer to specific questions and therefore might be more suitable
to assess maths anxiety across a wide range of age groups and abilities. However, it has not been
subject to intensive testing of its reliability or validity across a range of ages. Furthermore, this scale
does not show the relation between maths performance and maths anxiety which is almost

                                                                                                         10
universally observed using other measures (Thomas & Dowker, 2000). This might indicate that it is
not a valid measure of maths anxiety (Carey, Hill, et al., 2017).

Creation of the modified AMAS
Because of these problems with child maths anxiety scales, we decided to modify one of the most
commonly used adult maths anxiety scales, the AMAS, to be used with British children and
adolescents across our sample’s entire age range. The AMAS has consistently been shown to be valid
and reliable, despite its short (9 item) length. This includes several validations of translations of the
AMAS, suggesting that it might be robust to minor linguistic adaptations. Below is a table showing
items from the original and modified AMAS. Some of the original AMAS items measure anxiety
about maths tests (those marked with an asterisk). Others measure anxiety about learning maths
(Hopko et al., 2003). Our adaptations translated US-English to UK-English and ensured that items
would be age-appropriate for children across our sample. The scale has been used previously but not
validated (Zirk-Sadowski, Lamptey, Devine, Haggard, & Szucs, 2014).

Item                       Original AMAS                                          Modified AMAS
  1      Having to use the tables in the back of a math             Having to complete a worksheet by yourself
                             book
  2      Thinking about an upcoming math test 1 day                 Thinking about a maths test the day before
                            before*                                                 you take it
  3        Watching the teacher work an algebraic                    Watching the teacher work out a maths
                  equation on the blackboard                                  problem on the board
  4        Taking an examination in a math course*                             Taking a maths test
  5       Being given a homework assignment of many                   Being given maths homework with lots of
           difficult problems that is due the next class           difficult questions that you have to hand in the
                             meeting*                                                   next day
  6             Listening to a lecture in math class               Listening to the teacher talk for a long time in
                                                                                         maths
  7        Listening to another student explain a math             Listening to another child in your class explain
                              formula                                               a maths problem
  8           Being given a “pop” quiz in math class*                  Finding out that you are going to have a
                                                                       surprise maths quiz when you start your
                                                                                      maths lesson
  9           Starting a new chapter in a math book                          Starting a new topic in maths
Table 1 Items from the original and modified AMAS questionnaires

Assessment of the modified AMAS
First, we wished to investigate whether the modified AMAS is a reliable test. In order to do this, we
looked at whether scores on each item of the scale were related to scores on each other item of the
scale. We found that there was a very strong relationship between all items on the questionnaire,
suggesting that it reliably measures one construct. Secondly, we looked at the structure of the
modified AMAS, to see if it was the same as the original AMAS. In the original AMAS, items
measuring Learning maths anxiety were more related to one another than they were to those
measuring Evaluation maths anxiety, and vice versa. Thus it can be considered that the original

                                                                                                            11
AMAS consists of two subscales, measuring related but distinct forms of maths anxiety. The first
subscale (items 1, 3, 6, 7 and 9) measures anxiety about learning maths. The second subscale (items
2, 4, 5 and 8) measures anxiety about evaluation in maths (Carey, Hill, et al., 2017).

We used confirmatory factor analysis, which investigates how items from a questionnaire measure
“cluster” (i.e. which items are most interrelated to one another) in order to confirm that a measure
has a particular structure. This analysis showed us that the modified AMAS had the same underlying
structure as the original AMAS. That is, the modifications we made to the AMAS items did not
appear to change what was being measured by each item or how each item relates to each other
item. The structure of the mAMAS can be seen in Figure 1 (Carey, Hill, et al., 2017).

Figure 1 Path diagram showing how items from the mAMAS cluster onto two subscales, measuring Learning
mathematics anxiety and Evaluation mathematics anxiety

Whilst this was of great interest, it only confirms that the modified AMAS measures one construct
which can be subdivided into two sub-constructs. We were also concerned in showing that the
broad construct measured by the modified AMAS really was maths anxiety rather than something
else. Researchers in the past, for example, have suggested that maths anxiety might merely be one

                                                                                                        12
form of test anxiety (anxiety experienced in evaluative settings; Hembree, 1990). It is also possible
that, when working with children, the measure could merely assess how anxious the children felt in
general about events, behaviours and competence (general anxiety), rather than being specific to
emotions regarding mathematics (Carey, Hill, et al., 2017).

In order to assess whether the modified AMAS measured something specific to maths, we looked at
the relationship between results on the modified AMAS and results on the two other anxiety
measures we used, one of which looked at test anxiety and the other of which looked at general
anxiety. These measures – the Child Test Anxiety Scale (Wren & Benson, 2004) and the Revised
Children’s Manifest Anxiety Scale II (Reynolds & Richmond, 2012) – have been previously validated.
In order to see whether the modified AMAS measures something distinct from test and general
anxiety, we looked at how items from all three scales formed into clusters which were more related
to other items within the cluster than items outside of the cluster. We found that the scales
clustered into items measuring five different factors. One of these factors was best understood as
representing maths anxiety. This factor was measured by all items in the modified AMAS (Carey, Hill,
et al., 2017).

The remaining four factors which came out of this exploratory factor analysis were: test anxiety,
physical anxiety, off-task behaviours and social anxiety. Of particular interest was that two items
from the modified AMAS (“Thinking about a maths test the day before you take it” and “Taking a
maths test”) loaded more strongly onto the test anxiety factor than the maths anxiety factor. This
might suggest that whilst all items in the modified AMAS do measure maths anxiety (as they all
loaded onto the maths anxiety factor), some of them capture a mixture of maths and test anxiety
(Carey, Hill, et al., 2017).

Therefore, in this study we have shown three important properties of the modified AMAS:

    1. It is reliable – each item in the modified AMAS is strongly related to each other item,
         suggesting that it reliably measures one construct.
    2. It has good construct validity – its underlying structure mirrors that of the original AMAS,
         suggesting that both questionnaires measure maths anxiety.
    3.   It has divergent validity – items in the modified AMAS measure something different from
         what is measured by other anxiety scales.

Conclusions
These three properties are very important when assessing a questionnaire, and suggest that this is a
good instrument with which we can measure maths anxiety across a broad age range (at least 8-13
years). This could be of great practical interest to researchers and educational practitioners alike.
The short length of the modified AMAS means that it is very quick to administer and score, which is
useful to those who have limited time working with each child. Furthermore, having a measure
which can be used across a wide variety of ages is of great interest to those involved in assessing
many children of different ages, such as educational psychologists and teachers. For researchers it is

                                                                                                        13
also very interesting to use a questionnaire with the same structure as one of the most common
adult maths anxiety measures, as it may mean that we are able to compare results obtained from
child samples to those which we have from adults (Carey, Hill, et al., 2017).

Reference
Carey, E., Hill, F., Devine, A. and Szűcs, D. (2017). The Modified Abbreviated Math Anxiety Scale: A
Valid and Reliable Instrument for Use with Children. Frontiers in Psychology, 8.

Exploring the relationship between mathematics anxiety and performance

Before investigating the individuals in our own sample, we took an overview of the literature
concerning mathematics anxiety and maths performance. Research has long seen a relationship
between maths anxiety and performance in maths tests (Ashcraft & Krause, 2007; Devine, Fawcett,
Szűcs, & Dowker, 2012; Jansen et al., 2013; Zakaria, Zain, Ahmad, & Erlina, 2012). This relationship is
similar to that seen between other forms of anxiety (e.g. test anxiety) and test performance, and
consists of a small negative correlation (Mandler & Sarason, 1952). That is, as maths anxiety levels
increase, maths test performance somewhat decreases. Whilst this relationship is consistently
observed, there is a lack of consistency in the proposed direction of the relationship: that is, does
maths anxiety cause people’s performance in maths to decrease, or does having poorer
performance in maths cause people to become more anxious about the subject? The possible
directions of the relationship between anxiety and performance can be summarised with two
different theories: The Deficit Theory and the Deleterious Anxiety Model (Carey et al., 2015).

The Deficit Theory
The Deficit Theory suggests that people who start out with poorer maths performance are more
likely to develop anxiety about maths, as summarised in Figure 2 (Carey et al., 2015).

      Poor maths performance                                          Increased maths anxiety
     Figure 2 Image showing a causal relationship from poor mathematics performance to increased
     mathematics anxiety

For example, studies have suggested that children with mathematical learning disabilities such as
developmental dyscalculia (which causes reduced maths performance) have higher levels of
mathematics anxiety than children without mathematical learning disabilities (Passolunghi, 2011;
Rubinsten & Tannock, 2010). Longitudinal studies (studies which follow children over a longer period
of their development than most studies, which only provide a snapshot at a specific time point) also
suggest that decreased performance in mathematics might be linked to higher maths anxiety in the
following school year (Ma & Xu, 2004; Meece et al., 1990).

Furthermore, it has been suggested that adults with maths anxiety might have problems with basic
numerical processing (number sense), indicating that perhaps their performance was impaired at a
very early stage, before they developed maths anxiety (Maloney et al., 2011). Genetic evidence has

                                                                                                        14
also been found which indicates that some of the variation in maths anxiety can be explained by
genes which affect a person’s maths performance (Wang et al., 2014).

However, this research is not definitive. Whilst it might suggest that some individuals develop maths
anxiety as a result of poorer maths performance, not all individuals with maths anxiety have any
history of a performance deficit. Nor is it the case that all individuals with difficulties in maths go on
to develop maths anxiety. In addition, other research suggests that the link between maths anxiety
and performance can be driven in the other direction (Carey et al., 2015).

The Deleterious Anxiety Model
The Deleterious Anxiety Model suggests that the link between maths anxiety and maths
performance is driven by anxiety’s devastating consequences on learning and recalling maths skills,
as summarised in Figure 3 (Carey et al., 2015).
  High maths anxiety                                              Decreased maths performance

Figure 3 Image showing a causal relationship from high mathematics anxiety to decreased mathematics
performance

Maths anxiety may have an effect on maths performance at several different levels. Firstly, evidence
suggests that people with maths anxiety are less willing to engage with maths tasks at all. For
example, people with maths anxiety are less likely to enrol in maths classes (Hembree, 1990), and
have a tendency to answer questions quickly but inaccurately (perhaps due to trying to “escape” the
anxiety-inducing maths situation; Ashcraft & Faust, 1994). This suggests a tendency towards maths
avoidance in those with maths anxiety, which has a negative impact on both learning opportunities
and recall in tests (Carey et al., 2015).

Secondly, whilst individuals are engaged in maths tasks, maths anxiety might act to distract them
from what they are trying to learn or remember. The idea that anxiety could interfere with learning
and recall is known as ‘cognitive interference’ – anxiety generates distracting thoughts and
sensations which affect memory capacity. This idea is supported by evidence suggesting that those
with higher maths anxiety have poorer working memory (memory used to store, process and
manipulate information), and that those with maths anxiety do especially poorly in questions which
require a high level of working memory to solve (Ashcraft & Kirk, 2001; Ashcraft & Krause, 2007).

The idea is that maths anxiety reduces maths performance both by reducing engagement with
maths tasks and by making these maths tasks harder to solve by reducing working memory capacity.
There is some evidence that the relationship between maths anxiety and performance does operate
in this direction. For example, studies which elevate maths anxiety in specific individuals (for
example, by making women conscious of gender stereotypes about women being bad at maths) find
that this decreases maths performance (Galdi, Cadinu, & Tomasetto, 2013; Gerstenberg, Imhoff, &
Schmitt, 2012; Marx, Monroe, Cole, & Gilbert, 2013; Schmader, 2002; Seitchik, Jamieson, & Harkins,
2012; Spencer, Steele, & Quinn, 1999). Other studies have people do a task aimed to reduce maths
anxiety, and have observed an immediate performance increase (Park & Ramirez, 2014).

                                                                                                        15
Brain imaging data also suggests that anxiety might interfere with maths performance. For example,
there is evidence that when individuals with high levels of maths anxiety outperform those with
slightly lower levels, this happens alongside activation of the brain regions associated with cognitive
control of emotions. This suggests that those with increased emotional regulation are better able to
overcome maths anxiety (Lyons & Beilock, 2012).

The Reciprocal Theory
The mixture of evidence for each of the two theories suggests that in fact they might both play a
part in the relationship between maths anxiety and performance. That is, maths anxiety might cause
decreased performance and poorer performance might elicit maths anxiety, as summarised in Figure
4 (Carey et al., 2015).

    Increased maths anxiety                                        Decreased maths performance

  Figure 4 Diagram showing a reciprocal relationship between increased maths anxiety and decreased maths
  performance

We believe that a model like this one is best able to account for the mixture of data, which suggests
that the relationship between maths anxiety and maths performance operates in both directions.
Whilst few studies find explicit proof of a bidirectional relationship, this could be because the two
different directions are each best supported by different kinds of study. The effect of maths anxiety
on performance seems to most easily be observed in studies which change a person’s maths anxiety
(in the short term) and see if this affects their performance. On the other hand, longitudinal (long
term) studies find it easier to see the effect which performance has on future maths anxiety. In the
future, if single studies are to find evidence that the relationship between maths anxiety and maths
performance operates in both directions, it could be useful to carry out “mixed methods” research –
that is, research which uses multiple types of methods to investigate this relationship (Carey et al.,
2015).

This research on the relationship between maths anxiety and maths performance helped to guide
our interpretation of our own data. The idea that the relationship between maths performance and
maths anxiety is bidirectional suggests two things:

    1. Those with cognitive learning difficulties in mathematics, such as developmental dyscalculia,
         may experience higher levels of maths anxiety.
    2. Other factors, such as an individual’s predisposition towards anxiety generally might also act
         to raise maths anxiety.

                                                                                                           16
Reference
Carey, E., Hill, F., Devine, A. and Szücs, D. (2016). The Chicken or the Egg? The Direction of the
Relationship Between Mathematics Anxiety and Mathematics Performance. Frontiers in Psychology,
6.

Relationship between maths anxiety and developmental dyscalculia

Mathematics anxiety and developmental dyscalculia
We used data from phase 1 of our study to inform us about the relationship between maths anxiety
and developmental dyscalculia. Developmental dyscalculia is a difficulty in acquiring mathematical
skills, experienced by around 6% of children (Devine, Soltesz, Nobes, Goswami, & Szűcs, 2013).
Whereas maths anxiety is an emotional problem, developmental dyscalculia is a cognitive one
(Devine et al., 2017). Many different cognitive abilities have been linked to developmental
dyscalculia. For example, researchers have suggested that developmental dyscalculia could be linked
to problems representing magnitude of numbers, poor working memory (the ability to hold
information in mind whilst performing another task), difficulties with inhibition, poor spatial skills or
problems organising sounds (Devine et al., 2013). It is unclear which of these factors truly underlies
developmental dyscalculia, or if dyscalculia in fact represents a diverse range of mathematics
learning problems which have a variety of different factors at their root.

However, as we found in our review of the relationship between maths anxiety and maths
performance, some studies have found that children with cognitive maths problems, such as
developmental dyscalculia, are more likely to experience maths anxiety (Passolunghi, 2011;
Rubinsten & Tannock, 2010). We wanted to look at the prevalence of developmental dyscalculia and
maths anxiety across our large sample of primary and secondary students. We also wanted to look at
how gender differences differentially affect cognitive and emotional maths learning problems.

Gender differences in mathematics learning problems
The gender ratios reported in past studies of developmental dyscalculia are mixed. Some studies
show that developmental dyscalculia is present more in boys than girls (Barahmand, 2008; Reigosa-
Crespo et al., 2012; von Aster, 2000), others suggest that more girls than boys have dyscalculia
(Dirks, Spyer, van Lieshout, & de Sonneville, 2008; Hein, Bzufka, & Neumärker, 2000; Lambert &
Spinath, 2014; Landerl & Moll, 2010). Still further studies show no gender difference in dyscalculia
prevalence (Devine et al., 2013; Lewis, Hitch, & Walker, 1994).

Gender differences in maths anxiety studies are more consistent, with girls showing higher levels of
maths anxiety than boys in the vast majority of studies which found a gender difference (e.g. Devine
et al., 2012). Gender differences in maths anxiety are shown more in studies of adults and
secondary-school level children, indicating that the prevalence of maths anxiety in primary students
is less gender-dependent (Hill et al., 2016). Whilst there may not be any more girls than boys with
cognitive difficulties in maths, they appear to be more susceptible to emotional maths learning
problems (Devine et al., 2017).

                                                                                                       17
Why we looked at maths anxiety and dyscalculia
As discussed previously, researchers typically observe moderate negative correlations between
maths anxiety and maths performance, suggesting that those with slightly poorer performance in
maths are more likely to experience maths anxiety (Carey, Hill, Devine, & Szucs, 2016; Hembree,
1990). Various studies suggest that this relationship also holds in individuals with developmental
dyscalculia: these people are more likely to experience maths anxiety than their non-dyscalculic
counterparts (Passolunghi, 2011; Rubinsten & Tannock, 2010). However, whilst studies have shown
an increase in maths anxiety levels in dyscalculic individuals, no research has investigated the
prevalence of co-occurrence of maths anxiety and developmental dyscalculia (Devine et al., 2017).

Therefore, we decided to investigate the relationship between maths anxiety and developmental
dyscalculia in our large sample. First, we looked at the relationship between maths anxiety and
maths performance, not only in the whole sample but also in a subgroup who we identified as
having developmental dyscalculia. Secondly, we investigated the proportion of individuals who met
criteria for both maths anxiety and developmental dyscalculia, and whether there was a different
frequency of co-occurrence in girls and boys (Devine et al., 2017).

The relationship between maths anxiety and dyscalculia
Most people in our sample had low scores in the mAMAS (our maths anxiety questionnaire). As
mAMAS scores increase, the number of children who reported that maths anxiety level decreases.
We decided to define “high maths anxiety” as having a mAMAS score above the 90th percentile –
meaning that individuals were defined as having high maths anxiety if they were in the top 10% of
maths anxiety scores. This corresponded to a score of 30 on the mAMAS, meaning students with
high maths anxiety were, on average, selecting above “Moderate amount of anxiety” for each scale
item. We defined dyscalculia as a maths performance significantly below average (below 85, where
100 represents average performance) with a reading performance at or above the average range
(above 85, on the same scale). This way we found the students in our sample who had a selective
weakness in maths – i.e. they performed reasonably well in reading but not in maths (Devine et al.,
2017).

Whilst there was a negative relationship between maths anxiety and maths performance in the
sample as a whole, this relationship was not found in the dyscalculic children. This might just be
because their performance scores did not vary enough for the relationship to be seen. On the other
hand, it might indicate that once your maths performance is below a certain level, it doesn’t really
matter how far below that level you are; you are equally likely to experience maths anxiety
regardless. We found that the likelihood of experiencing high maths anxiety was 22% for the
dyscalculic children, compared with 10% in the sample as a whole. This confers a statistically
significantly increased risk of developing maths anxiety in children with dyscalculia than those
without (Devine et al., 2017).

However, whilst maths anxiety risk is much higher in those with dyscalculia than those without,
dyscalculic children still make up an overall small proportion of all of those with high maths anxiety.

                                                                                                       18
In those with high maths anxiety, only 11% fell into the developmental dyscalculia group and 12%
had global low performance (below 85 in both maths and reading). The vast majority (77%) of
children with high maths anxiety had typical or above-typical maths performance. Gender appeared
to play a significant role in the development of maths anxiety in dyscalculic children. Whereas boys
with dyscalculia had a low chance of having co-occurring high maths anxiety, girls with dyscalculia
were at a much greater risk of having elevated maths anxiety. This fits with what we already knew:
girls are more susceptible to maths anxiety than boys. This research shows a large degree of
dissociation between cognitive and emotional maths learning problems, suggesting that each may
require a different intervention (Devine et al., 2017).

Reference
Devine, A., Hill, F., Carey, E., & Szűcs, D. (2017). Cognitive and Emotional Math Problems Largely
Dissociate: Prevalence of Developmental Dyscalculia and Mathematics Anxiety. Journal of
Educational Psychology. https://doi.org/10.1037/edu0000222

Maths anxiety: Gender differences, developmental change and anxiety specificity

In collaboration with Italian researchers, we investigated gender differences, developmental change
and specificity of maths anxiety in Italian primary and secondary school students. This enables us to
investigate whether any findings are specific to the UK, or if they are more general.

Previous findings
Previous research in this area has shown that girls tend to have higher levels of anxiety about maths
than boys, despite generally showing similar levels of maths performance (Devine et al., 2012). This
may result from a higher disposition towards anxiety in general, as girls also tend to show higher
scores on other forms of anxiety (Lewinsohn, Gotlib, Lewinsohn, Seeley, & Allen, 1998).
Alternatively, it could relate to gendered attitudes held about mathematics competence and
suitability (Tomasetto et al., 2011).

Findings related to developmental changes in anxiety are less clear-cut. It is possible that the link
between mathematics anxiety and performance does not emerge until later in school (Dowker,
2005; Thomas & Dowker, 2000). Additionally, whilst maths anxiety is by definition a specific form of
anxiety experienced towards mathematics, research has consistently shown individuals with maths
anxiety to also show higher levels of test and general anxiety (Punaro & Reeve, 2012; Wang et al.,
2014). This calls into question whether maths anxiety is truly a maths-specific anxiety form, or just
one manifestation of more generalised anxieties (Hill et al., 2016).

In this research we aimed to further explore gender differences in maths anxiety; investigate more
about developmental change in the relationship between maths anxiety and maths performance
and investigate the specificity of maths anxiety by looking at how it relates to reading performance
and general anxiety levels.

                                                                                                        19
Participants and measures
We worked with around 1000 Italian students, 639 attending primary school with an average age of
9 years, and 342 attending secondary school with an average age of 13 years. Children completed
arithmetic and reading comprehension tests, alongside questionnaires assessing mathematics
anxiety and general anxiety levels (Hill et al., 2016).

Our findings: Gender
As can be observed in Figure 5, girls had higher levels of both mathematics and general anxiety. This
was the case regardless of whether considering primary school students, secondary school students,
or both. This confirms previous research indicating that girls report higher anxiety than boys, both in
mathematics and in general. Findings by level of schooling were not significantly different for the
AMAS or RCMAS-2, though secondary school students scored higher in the AMAS and primary school
students scored higher in the RCMAS-2 (Hill et al., 2016).

Figure 5 Average maths anxiety (AMAS) and general anxiety (RCMAS-2) scores in different groups, split by a) gender, b)
school level, c) gender within primary students and d) gender within secondary students.

Our findings: Anxiety specificity
Students’ maths anxiety and general anxiety scores were correlated – this means that as student
maths anxiety increases, general anxiety scores also tend to increase. This is graphically displayed in

                                                                                                                    20
Figure 6 below. This shows that whilst those with higher general anxiety tend to have higher maths
anxiety, there is still a lot of variability in maths anxiety reports for any given general anxiety score.
That means that whilst these anxiety forms are related, one cannot be entirely explained by the
other. This correlation persisted regardless of gender and level of schooling (Hill et al., 2016).

Figure 6 Scatter plot of individuals’ general anxiety (RCMAS-2) and maths anxiety (AMAS) scores.

Our findings: Developmental changes
Researchers have shown a consistent relationship between maths anxiety and maths performance in
adolescents and adults. However, some have not found the same relationship in younger children.
Our research supports this – we found much stronger relationships between maths anxiety and
performance in secondary school boys and girls than in primary school. Whilst primary school girls
did show some relationship between maths anxiety and maths performance (higher maths anxiety
resulting in slightly poorer performance), primary school boys showed no such relationship.
Furthermore, once we took into account levels of general anxiety, there was no remaining
relationship between maths anxiety and performance in primary school girls or boys (Hill et al.,
2016).

                                                                                                         21
You can also read